Model comparison
Grok 4.20 (Non-Reasoning) vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.0 to 48.5 on the Noometry Index. Grok 4.20 (Non-Reasoning) costs 1.9× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
Alibaba (Qwen)Summary
- They share 31 benchmarks with published results for both. Grok 4.20 (Non-Reasoning) scores higher in 0 categories and Qwen3.8 Max in 10 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3.8 Max leads 73.1 to 48.0.
- The biggest single-benchmark swing is ProofBench: 14% for Grok 4.20 (Non-Reasoning) and 58% for Qwen3.8 Max.
- Grok 4.20 (Non-Reasoning) is cheaper at $1.25 / $2.50 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
Side by side
| Grok 4.20 (Non-Reasoning) | Qwen3.8 Max | |
|---|---|---|
| Provider | xAI | Alibaba (Qwen) |
| Noometry Index | 48.5 | 56.0 |
| Released | 2026-02-17 | 2026-08-02 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1M |
| Max output | 30K | 131K |
| Input $ / M tokens | $1.25 | $2 |
| Output $ / M tokens | $2.50 | $6 |
| Results tracked | 46 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 41.8 (#104), Qwen3.8 Max: 52.3 (#30)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena WebDev | 1374 | 1671 |
| LMArena Coding | 1458 | 1499 |
| DeepSWE | — | 57.5% |
| FrontierSWE | — | 15.8% |
| SciCode | — | 52.9% |
| WeirdML | 52.3% | — |
| ALE-Bench | 1,150 | — |
Agentic & Tool Use Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 34.3 (#43), Qwen3.8 Max: 42.6 (#17)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| τ²-bench Banking | 18% | 55.1% |
| Terminal-Bench | 57.3% | — |
| GDP.pdf | — | 23.2% |
| LMArena Search | 1189 | — |
| Vending-Bench 2 | 4,663 | — |
Reasoning Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 52.2 (#30), Qwen3.8 Max: 53.4 (#26)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| NYT Connections (extended) | 85.4% | 88.3% |
| Chess Puzzles | 24% | 40% |
| LMArena Hard Prompts | 1451 | 1493 |
| DTBench | 90.1% | 92% |
| LMCA | 38.7% | 46.2% |
| Epoch Capabilities Index | 152.01 | 156.69 |
| ARC-AGI-2 | 65.1% | — |
| Kagi LLM Benchmark | 75% | — |
| ARC-AGI-1 | 89.5% | — |
| CritPt | — | 20% |
| Thematic Generalization | 63.8% | — |
| Mystery Game Puzzles | — | 38% |
| ForecastBench | 61.4 | — |
Math Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 48.0 (#54), Qwen3.8 Max: 73.1 (#15)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | 44.9% | 74.7% |
| FrontierMath Tier 4 | 17.1% | 46.3% |
| OTIS Mock AIME 2024-2025 | 92.2% | 100% |
| ProofBench | 14% | 58% |
| LMArena Math | 1455 | 1498 |
Knowledge Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 52.8 (#54), Qwen3.8 Max: 61.7 (#25)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 89.3% | 92.7% |
| SimpleQA Verified | 30.2% | 47.3% |
| LMArena Expert | 1440 | 1504 |
Multimodal Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 33.2 (#88), Qwen3.8 Max: 36.9 (#70)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | 1263 | 1314 |
| Blueprint-Bench 2 | 0% | — |
| Furniture Assembly | — | 20% |
| LMArena Document | 1416 | — |
Multilingual Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 54.7 (#38), Qwen3.8 Max: 56.6 (#18)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1440 | 1468 |
| LMArena Chinese | 1481 | 1537 |
| LMArena French | 1477 | 1502 |
| LMArena German | 1463 | 1489 |
| LMArena Japanese | 1450 | 1466 |
| LMArena Korean | 1417 | 1467 |
| LMArena Russian | 1457 | 1478 |
| LMArena Spanish | 1442 | 1495 |
Instruction Following Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 75.0 (#79), Qwen3.8 Max: 77.7 (#19)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1420 | 1478 |
Long Context Too close to call
Grok 4.20 (Non-Reasoning): 45.4 (#33), Qwen3.8 Max: 45.5 (#30)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1436 | 1486 |
| CL-bench | 22.2% | — |
| CL-bench Life | 11.9% | — |
Writing & Preference Qwen3.8 Max leads
Grok 4.20 (Non-Reasoning): 65.9 (#42), Qwen3.8 Max: 67.0 (#31)
| Benchmark | Grok 4.20 (Non-Reasoning) | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1451 | 1480 |
| LMArena Creative Writing | 1439 | 1478 |
| LMArena Multi-Turn | 1456 | 1487 |
| EQ-Bench Creative Writing | 1574 | — |
Frequently asked questions
Is Grok 4.20 (Non-Reasoning) better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.0 to 48.5 on the Noometry Index. Grok 4.20 (Non-Reasoning) costs 1.9× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, Grok 4.20 (Non-Reasoning) or Qwen3.8 Max?
Grok 4.20 (Non-Reasoning) is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is Grok 4.20 (Non-Reasoning) or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 52.3 versus 41.8 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do Grok 4.20 (Non-Reasoning) and Qwen3.8 Max share?
31 benchmarks have published results for both models. Grok 4.20 (Non-Reasoning) has 46 scored results on Noometry and Qwen3.8 Max has 38.